willem a. landman francois engelbrecht ruth park
TRANSCRIPT
THE CCAM AS OPERATIONAL SEASONAL
FORECAST SYSTEMWillem A. Landman
Francois Engelbrecht
Ruth Park
Building an optimized CCAM seasonal forecast system
Objective: to produce skilful seasonal forecasts at lead-times up to 6 months
Operational seasonal forecast development is a function of the ability of the next “best” system to outscore the current base-line skillAfter AGCMs, CGCMs is theoretically the next “best”
system (challenge for WG3) Optimal systems have the best chance to
capture important modes of variability and their link to SADC’s seasonal-to-interannual variability A large AMIP and hindcast data set will be
available for this purpose: Challenge for WG1
Old operational approach
Verification of old system: Limpopo(also a challenge for WG2)
WG3: To improve on drought forecasting
Streamflow forecast skill (DJF)
850 hPa CCAM simulations downscaled to streamflow
New operational approach
NCEP/GFS
Model Output StatisticsAtmospheric ICs
Boundary ConditionsResolution ~200km
Should we direct (some of) our focus to the southern/mid-latitudinal ocean?Challenge for WG3?
AUG ICs
SEP ICs
OCT ICs
NOV ICs
Predicted Subtropical Dipole Modes during 2010/11
Inclusion of SINTEX-F forecasts in the MM should improve skill
Imminent development
AMIP1979 to 20086 ensemble members
Hindcasts with predicted SSTs1982 to 201010 ensemble members
Verification statisticsSVSLRF
Applying forecasts toStreamflowMaize yield
What about the land? Land surface conditions may modulate the response of the
atmospheric circulation to SST anomalies Agents of climate memory at the land surface
Soil moisture Snow cover State of vegetation
“If the general circulation alone determines local anomalies, and SST determines the general circulation, then there is little hope for enhancing prediction during boreal summer by improved land surface representation”
Is there latent predictability over a land region to be harvested from the land surface state? If so, would it supersede SST influences?
CCAM will be integrated, coupled to the dynamic land-surface model CABLE, in an attempt to investigate the relative role of the land-surface in forcing seasonal rainfall and temperature anomalies over southern Africa
ENSEMBLES
Strong anthropogenically forced warming trends have been observed over southern Africa and are projected to continue to rise, consequently justifying the investigation into how the annual update of greenhouse gas (GHG) concentrations in a global model may affect seasonal forecast performance over the region.
1901-2002
Future plans: SATREPS-2 ??
APPLICATIONS
Medium- to extended-range (beyond 3 days)
Seasonal forecast system (current
SATREPS)
1-3 years; decadal
Maize yieldRiver flow
Diseases
Livestock
Tornado Sunday
Hundreds of homes were destroyed in Ficksburg in the Free State. Another tornado hit the East Rand and caused extensive damage to Duduza, near Nigel. Two children died.
Final comments… Optimal AGCM configuration will benefit from sensitivity
studies using AMIP (to determine, for example, Cu scheme, etc.) [WG3]
Resources should continue to be directed towards AGCM optimization [WG3]about ½ resources required compared to CGCMs – higher resolution,
bigger ensemble SA modellers focussing on CGCM development/use – must outscore
baseline to justify effort More effort should be directed towards analysing AGCM
hindcast/AMIP data to understand processes [WG1] Hindcast global SST set: 28 years, 6 months lead-time, AND
operational SST forecasts available from CSIR FTP site (UCT-CSAG already using it for AGCM predictions, soon at SAWS and at CSIR) [WG3/4]
Strong emphasis on applications modelling [WG4]